[<<] Industrie Toulouse

I've been compiling notes about the Lisa, Mac OS, Windows, OS/2, Mac OS X, Cairo, and more, in the recent week. I've been doing this in order to write some entries about where OS design has been, is going, may be going, should have been, could have been, etc. A lot of what I've compiled so far has been written up already, either by me or by others. I wanted to share a few of these:

John Siracusa, "About the Finder..."

Published on Ars Technica during the year of Jaguar, About the Finder gives an excellent history about the strengths of the classic Mac OS Finder that have been lost in Mac OS X. He offers up a proposal to address the issues in the Mac OS X finder, which is to visually separate regular mode with a brushed metal single window "browser" mode. Which is not too dissimilar to what the new Finder offers in Mac OS X 10.3, although the true spatial features (one and only one window may ever represent a folder (with the exception of search results)) are still lacking. Apple has at least made the single window mode in Panther much more useful, and heavily distinguished from plain windows.

Siracusa also hopes for a metadata based Finder with search folders. I hope these aren't too far off. iTunes is driving a lot of Apple's interface designs (for better or for worse), and iTunes introduced the concept of "Smart Playlists" a couple of revisions ago. Smart Playlists are live searches - 25 most played songs, most recently added, etc. The concept has also shown up in Apple's new Xcode IDE, with "Smart Groups" letting you have virtual folders showing all implementation files in a project, or all jpg's, or anything else, executed as live queries when you visit them. The fast search field in iTunes has also been showing up everywhere lately, including the new Finder wherein search results start showing up as you're typing (as fast as it can find them, which depends of course on scope), and Apple has included this in their available frameworks as SearchKit, a search engine system other developers can make use of to provide similar functionality in their own applications and tools. So, search is becoming quite important for Apple and they seem to be packaging it well. Here's hoping that "Smart Folders" or something similar show up in Mac OS X 10.4. I'd love to have a folder that would always find files I've marked as "Critical."

I'm hoping that Siracusa is the one to write Ars Technica's review of Panther. I'm very interested in his take on the new version of the Finder. I doubt it will make him completely happy, but I hope he sees it as steps in the right direction.

Matthew Thomas, "When Good Interfaces Go Crufty"

A great post about the user interface cruft we've picked up over the years. I don't agree with everything he says, but there are a lot of great points made here. One that I found entertaining is the continued reliance a lot of Microsoft software (and apparently Windows software in general) has on paths, instead of file ID's and inodes. Sometimes the mere changing of a folder name can cause significant breakage. I hope this situation is better today.

When Microsoft brought Word 6.0 to the Mac, they decided to operate under the assumption that Mac users wanted Word to behave exactly the same on the Mac as it does on Windows. And I believe this was Microsofts chance to use their cross platform MFC libraries. What this lead to was a very fragile version of Office. It's not uncommon for Mac users to rename their hard drive (we're not bound by single characters). But doing so would cause Word to completely stop working because it lost...well, who knows what it lost. But it lost it. Microsoft improved on this greatly with the Mac editions of Office 98 and Office 2001, which had the concept of being "self repairing." If an Office application started up and couldn't find support libraries in the System folder, it would launch a program to effectively do an automatic re-install of those files and then start up.

There are many other elements of cruft active today, some of which come from the restrictions that the original Mac OS had that its older sibling, the Lisa Office System, did not (there was no 'Quit' in Lisa, there was no "open/save" dialog boxes, there was no "file -> new" menu item). Many others come from Windows.

Wilfredo Sánchez, "The Challenges of Integrating the Unix and Mac OS Environments"

This paper, presented at USENIX 2000, details just how significant of a feat Mac OS X is. Unix and the Mac OS were engineered for very different purposes and with very different expectations as to how they would be used. Bringing those two very different sides together was no small feat. This was not like NeXTStep, which was built for Unix from day one. This was Apple having to maintain as much of their user bases expectations as possible while having to move this very single-user oriented operating system and its APIs and file systems to work on Unix without driving their standard base away. A very interesting read, and I think it's a good insight into why Mac OS X took so long to build and also why the first releases were a little rocky. I think they've done an excellent job.

Industrie Toulouse OS (de)Evolution Category

In the posts that I've brought into this category there are some reactions to "When Good Interfaces go Crufty," some looks back at the Lisa Office System and some early Lisa ideas that have shown up in Mac OS X, and a quick look back at Cairo - the version of Windows NT with the OFS (which is finally showing up in Longhorn as WinFS)